Urban congestion and accidents are increasingly frequent. In particular, traffic congestion at intersections is severe. If traffic flows cannot be effectively channeled in real time, road congestion will spread, and the efficiency of intersections will be reduced, causing great time and economic losses to the public. Video vehicle detection technology is to install video capture equipment on roads with complicated conditions or easy to form congestion, detect the number, speed and queue length of passing cars, and transmit the collected data back to the server center through wired or wireless network for processing. Technology, through the real-time acquisition of traffic parameters, dynamic traffic signal control can be carried out to achieve effective law of traffic flow, and to minimize traffic congestion.
At present, the signal control methods mainly include timing control, multi-time control, induction control, and adaptive control. The traditional model algorithm is too rigid to set the threshold according to the change of a certain traffic parameter to optimize the signal, which will cause the system to misjudge the state.